Skip to content

fix: standardize bullet point formatting in research interests section #20

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
RonghaiHe wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix: standardize bullet point formatting in research interests section #20

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions profiles/description.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,16 +1,16 @@
I am a researcher focused on AI Agent Safety. My core research interests span the following areas:

1. Agent Safety — Safety of LLM-based autonomous agents, including unsafe tool use, unintended side effects, goal misgeneralization, out-of-distribution behavior, and safe exploration in agentic workflows.
1. Agent Safety - Safety of LLM-based autonomous agents, including unsafe tool use, unintended side effects, goal misgeneralization, out-of-distribution behavior, and safe exploration in agentic workflows.

2. Hallucination — Factual errors, confabulation, and grounding failures in LLMs and agents; detection, mitigation, and evaluation of hallucination; retrieval-augmented generation as a countermeasure.
2. Hallucination - Factual errors, confabulation, and grounding failures in LLMs and agents; detection, mitigation, and evaluation of hallucination; retrieval-augmented generation as a countermeasure.

3. Misalignment — Value misalignment, reward hacking, specification gaming, Goodhart's Law, and failures of RLHF/RLAIF; scalable oversight, weak-to-strong generalization, and alignment evaluation.
3. Misalignment - Value misalignment, reward hacking, specification gaming, Goodhart's Law, and failures of RLHF/RLAIF; scalable oversight, weak-to-strong generalization, and alignment evaluation.

4. Deception — LLM deceptive behavior, sycophancy, sandbagging, hidden reasoning, strategic manipulation, and dishonesty in agents; detecting and eliciting deceptive behavior.
4. Deception - LLM deceptive behavior, sycophancy, sandbagging, hidden reasoning, strategic manipulation, and dishonesty in agents; detecting and eliciting deceptive behavior.

5. Codex / Code Agents — Safety of code-generating models and autonomous coding agents (e.g., OpenHands, SWE-agent, Codex); vulnerabilities introduced by agent-generated code, backdoors, and code-level attacks.
5. Codex / Code Agents - Safety of code-generating models and autonomous coding agents (e.g., OpenHands, SWE-agent, Codex); vulnerabilities introduced by agent-generated code, backdoors, and code-level attacks.

6. Prompt Optimization & Attacks — Adversarial prompts, jailbreaks, prompt injection, automated red-teaming, and prompt optimization methods (e.g., APE, GCG, AutoDAN); defenses and robustness to prompt-level attacks.
6. Prompt Optimization & Attacks - Adversarial prompts, jailbreaks, prompt injection, automated red-teaming, and prompt optimization methods (e.g., APE, GCG, AutoDAN); defenses and robustness to prompt-level attacks.

I am NOT interested in: computer vision safety, hardware security, network intrusion detection, classical ML robustness unrelated to LLMs.

Expand Down